The ACD Method

[gmst]

Quote from mdl060374:


One thing I am noticing, is that alot of the time a large relative (volatile) OR, leads to a larger intraday move, and vice versa.

The reason I bring this up is that awhile back, I think an idea that was being kicked around was that looking for small intraday ratios perhaps led to better setups. (small being a ratio of < .20 or around there)

From what I have seen so far, small OR's usually lead to little action, and volatile/sloppy OR's lead to bigger intraday runs. So it seems kind of the reverse than what was suspected.

I discovered the same thing in a different context though for ES. So, yes, I can confirm that what you say is correct for ES. Haven't done any testing on stocks - so cant say for them.

Yes, it is counter-intuitive. Actually, I was looking for first 60 min range of < 5 points for ES, but all the profits were in range > 7 points. For < 5 points, my strategy was negative. LOL

 

[Quon]

Quote from mfbreakout:

Best way to see if wide or, narrow or etc.. will materialize is keeping track of 30 days number line along with daily pivot range, 3 day rolling pivot range etc..

Like Maverick74 said, it's all relative.


If anyone knows a programmer who can write a program for me to keep track of 30 days number line along with a up/ a down levels etc.. for stocks, futures etc.. please let me know.

Do you really want to automate the 30 day number line though? Think about it. Instrument trades down early in the morning and comes within 3 ticks of your A down level. It doesn't penetrate the level however, and turns and runs to the A up, only to come off and settle 1 tick above the OR.

Now, a coded system is going to call that a +2 day, but I'd probably call it a +1 day. See what I mean? Let me break it down:

The "failed A down" isn't a failure to a coded system because it didn't touch/penetrate the threshold (your a down level). Does that mean that missing it by 3 ticks is not a failed A down? No, just means that a system without the ability to judge is going to say it is not a failed A down.

Next, the "confirmed A up" kinda sucked. It settled above the OR, so technically it should get a +2 right? In my opinion though, it sucked, and it gets nada to me. Might as well have closed in the OR.

So computer's count: +2 (confirmed A up, settles above OR)

My count: +1 (failed A down, settles "in" the OR)

 

[Shanb]

Quote from mdl060374:

I was going to ask about this, and these examples give me a good chance to. (they had large OR)

I have been looking at statistics on equities concerning OR/DailyATR ratios, and behaviors.

One thing I am noticing, is that alot of the time a large relative (volatile) OR, leads to a larger intraday move, and vice versa.

The reason I bring this up is that awhile back, I think an idea that was being kicked around was that looking for small intraday ratios perhaps led to better setups. (small being a ratio of < .20 or around there)

From what I have seen so far, small OR's usually lead to little action, and volatile/sloppy OR's lead to bigger intraday runs. So it seems kind of the reverse than what was suspected.

Can anyone comment on this or add any insight? B/c I have been looking to stay away from larger than usual OR's, but in hindsight it seems like those were the best opportunities, when I reivew each days activity on stocks.

(The only idea I can come up with is that finding small intraday OR's can be good candidates for moves that catch people off guard. Or perhaps failures off of an A up/down, to fade for a favorable R/R trade.)

How are you "looking at statistics"? Are you running it on a set of data or manually going back through? What instruments and how how far back is the data?

 

[Shanb]

John,

You were talking about finding "tells" a couple pages back. What are you looking for in a tell?

Are you looking for the strongest products to confirm first on a breakout and the weakest on a breakdown? This is one that I think would make some sense...i'm sure there are other things. Care to share?

 

[Maverick74]

Quote from Shanb:

John,

You were talking about finding "tells" a couple pages back. What are you looking for in a tell?

Are you looking for the strongest products to confirm first on a breakout and the weakest on a breakdown? This is one that I think would make some sense...i'm sure there are other things. Care to share?

To me, tells are events that are happening that will define the next move up or down. Seeing the strength in the chip stocks was huge. I haven't see them this strong since 1999. Bonds breaking down was a tell because over the last few weeks remember, they were rallying with the ES. They were holding bids even in weak days. Today they got hit pretty hard. Copper. Copper is really really getting strong. Copper is part of the China trade. If the economy picks up in China, it will lift everything. AAPL making new all time highs. The Aussie Dollar is getting really strong showing the strength in the metals. The XHB is on fire, those are the basic material stocks that rally ahead of a strong recovery. All these things are breaking out.

 

[mfbreakout]

Quote from Quon:

Do you really want to automate the 30 day number line though? Think about it. Instrument trades down early in the morning and comes within 3 ticks of your A down level. It doesn't penetrate the level however, and turns and runs to the A up, only to come off and settle 1 tick above the OR.

Now, a coded system is going to call that a +2 day, but I'd probably call it a +1 day. See what I mean? Let me break it down:

The "failed A down" isn't a failure to a coded system because it didn't touch/penetrate the threshold (your a down level). Does that mean that missing it by 3 ticks is not a failed A down? No, just means that a system without the ability to judge is going to say it is not a failed A down.

Next, the "confirmed A up" kinda sucked. It settled above the OR, so technically it should get a +2 right? In my opinion though, it sucked, and it gets nada to me. Might as well have closed in the OR.

So computer's count: +2 (confirmed A up, settles above OR)

My count: +1 (failed A down, settles "in" the OR)

MF subscription service gives 30 days number line for lot of futures, so i am set in terms of futures.

I am trying to get 30 days number line for stocks as i find it difficult to keep track of stocks with daily, weekly, monthly A up/ A down levels.

Now, a stock will only get +ve number if it passes TIME filter test. Just touching A up or A level does not mean a thing unless it stays above A up level for 1/2 of OR- as an example.

I am not sure without automation how one can keep track of lets say 30 stocks basket.

 

[Shanb]

Quote from Maverick74:

To me, tells are events that are happening that will define the next move up or down. Seeing the strength in the chip stocks was huge. I haven't see them this strong since 1999. Bonds breaking down was a tell because over the last few weeks remember, they were rallying with the ES. They were holding bids even in weak days. Today they got hit pretty hard. Copper. Copper is really really getting strong. Copper is part of the China trade. If the economy picks up in China, it will lift everything. AAPL making new all time highs. The Aussie Dollar is getting really strong showing the strength in the metals. The XHB is on fire, those are the basic material stocks that rally ahead of a strong recovery. All these things are breaking out.

I don't know if you remember but we talked about a feel for the markets at one point in time. I've noticed that when one has been glued into the markets for a certain period of time little things tend to stick out and its easier to connect to the whole picture. There's no A-B-C, alot of is just experience and a good feel for things. You obviously have a good amount of these

If one is doing this part-time it is really hard to develop this kind of feel. Which I'm sure helps you be the consistent trader that you are. This just gets me excited because it justifies the almost obsession this job can bring sometimes lol.

 

[Maverick74]

Quote from Shanb:

I don't know if you remember but we talked about a feel for the markets at one point in time. I've noticed that when one has been glued into the markets for a certain period of time little things tend to stick out and its easier to connect to the whole picture. There's no A-B-C, alot of is just experience and a good feel for things. You obviously have a good amount of these

If one is doing this part-time it is really hard to develop this kind of feel. Which I'm sure helps you be the consistent trader that you are. This just gets me excited because it justifies the almost obsession this job can bring sometimes lol.

Yeah feel can go away really fast. If I miss a single day of trading, I feel completely lost when I come back. I talked about this a few weeks back when I mentioned I put more then 60 hours a week into this and some people asked..."doing what?" LOL. Watching the market! I don't how one could just casually glance at charts and news every now and then and ever expect to get any kind of feel.
'

 

[Maverick74]

BTW, today. Mega tell in GS. They come out with numbers and are light on revenues and the stock is the strongest it's been in 6 months. That boys is called a tell! Bad news/good action!

 

[mdl060374]

Quote from Shanb:

How are you "looking at statistics"? Are you running it on a set of data or manually going back through? What instruments and how how far back is the data?

First thing I did was reduce the universe of stocks to a manageable number:

$15-100, DailyATR(20) > .75, averagevolume(30) >1,500,000

THen I had an intraday indicator created for TS that had inputs available for OR timeframe, # of days back to look at, and other criteria.

I used anywhere from 75-100 trading days as a lookback period (anymore isnt relative from a daytrding perspective I think. Correct me if I am wrong. The idea is to have a large enough sample size, but not too far back where its looking at environments that arent relative to what is happening recently)

I then had it tell me:

1) how many of those days was the OR/dailyATR ratio (an input is there to specify how many days for the atr is used)
less than .25 (using .25 as an example. its another input)

2) for each day that fulfilled THAT criteria, what % of THOSE days did the stock gain OR lose x% at some point in the day.

(I later changed #2 to not work on a % move comparison, but on a % of that denominator as a move.. (ie open + 65% of the daily ATR, and open - 65% of the ATR.

Regardless of the crtieria you determine to be a comparison for a"move" it was giving me an idea of what had happened behaviorally with stocks and the OR.

What I found was this. With an OR of 4 minutes, you obviously had a higher occurence of a tighter range (obviously) than 8 minutes. But you usually when that happened, 30-45% of the time it reached 3.5% - 4% intraday move..

(and this was after sorting results to most favorable, and focusing on that small portion)

Using 65% of the daily ATR, it was even less. The reason I stopped using the % move is that often % moves were too close to my "A levels", so I thought using a % of the ATR might make more sense.

After reviewing this for a bit, it does seem like low OR/ATR ratios arent that useful for finding moves.

However, I do think that tight OR's DO seem helpful in keeping a favorable risk reward setup. So if that is the intent, then yes it can be useful. I do believe that this info probably be used to find a strategy that works around keeping a favorable R/R, I just havent come up with it yet...

I am in the process now of coming up with a 30 day numberline, but dont know how feasible it is.. I have to talk to the programmer I know b/c coming up with that is way over my head from a programming perspective..

I am also trying to incorporate weekly, and monthly levels into my planning, as well as comparing how certain stocks are behaving relative to the overall market. Both things people on this thread (mainly Mav) have mentioned. I am using both 10 day relative strength, and intraday RS vs the S&P for that.

My biggest problem is that I am trying to work on many variables at once. I realize this lol.